中文摘要：

针对掌纹图像在进行非接触式采集时易产生离焦模糊图像,从而导致系统识别性能降低的问题,提出了基于VO图像分解模型的模糊掌纹识别方法.首先建立高斯离焦退化模型模拟图像的模糊过程,通过对模糊原理的研究与分析得出在图像模糊过程中存在稳定的特征,这种特征从图像层次结构划分可表现为图像的结构层;然后使用VO图像分解模型得到模糊掌纹的结构层图像;为了提高特征的可区分性,采用分块的梯度方向直方图提取结构层中的稳定特征;最后采用归一化相关性分类器度量特征间的相似度.在清晰PolyU掌纹库和模糊PolyU掌纹库上进行测试的实验结果表明,该方法在不同掌纹库上获得较优的识别精度,且识别结果较为稳定;在模糊PolyU掌纹库中的等错误率最小可达0.3091％,优于传统高性能掌纹识别方法;且进行一次身份辨识的时间小于1.3s,满足实时性要求,表明了该方法的有效性和优越性.

英文摘要：

In order to solve the problem that during the collection of palmprint images by using non- contact device, it is easy to generate the defocus blurred image and degrade the performance of the recognition system; a blurred palmprint recognition method based on VO decomposition was proposed. Firstly, Gaussian defocus degradation model was established for simulating the image blur. There exist stable features in the process of image blurring by analyzing the blur theory. Such stable feature was considered as the structure layer of image from the perspective of image layering. Thereby, the structure layer of blurred image was obtained by using VO decomposition model. We used the blocked histogram of oriented gradient to extract the stable features from the structure layer in order to improve the distinguishability of feature. Finally, normalized correlation classifier was used to measure the similarity of palmprint. The experimental results show that the recognition accuracy of the proposed method is superior and stable in the PolyU palmprint database and the Blurred-PolyU palmprint database, moreover, the equal error rate （EER. 0. 309 1%） of the proposed method is lower than the classical high-performance algorithms in Blurred-PolyU palmprint database. The identification of time is less 1.3 s at a time, which meets the real-time requirement. The effectiveness and real-time of the proposed method is in this paper verified.